Block has released Buzz 1.0, a workspace where people and AI agents share the same chat rooms, and Jack Dorsey announced it on X at 02:38 UTC on 11 November as “the most flexible team workspace out there”. Buzz’s pricing page offers the open-source software free to self-host, and its GitHub repositories carry the Apache 2.0 license.
The hosted version charges $0.65 per vCPU-hour of compute for remote agents and $0.15 per GB-month of data, after 100 GB a month included, with no monthly minimum, the page says. One remote agent on one vCPU, left running for a 730-hour month, would cost $474.50 in compute, by our arithmetic.
Business Pill · Multi-Agent Systems
An evergreen Business Engineer explainer on multi-agent systems: one agent or a team. It teaches the general idea only and does not discuss Buzz.
The key insight: As we read it, Block is giving away the workspace and charging for the work: the Buzz software is free to self-host under Apache 2.0, while the hosted plan bills compute at $0.65 per vCPU-hour, so one always-on agent on one vCPU costs $474.50 a month by our arithmetic.
The Buzz 1.0.0 release was published on GitHub at 09:40 UTC on 10 November in Block’s block/buzz-app repository, with builds for Apple Silicon and Intel macOS (signed and notarized), Windows x64 (an unsigned installer) and Linux. Dorsey’s post described it as “for humans and agents, model agnostic, local or cloud, and infinitely extendable”.
The app’s README lists four agent harnesses: Claude, which needs a Claude account, Codex, which needs a ChatGPT or OpenAI account, Goose and Pi. It says agents run on your computer while Buzz is open. When we read it on 11 November, the same README still said “Buzz 1.0 is in beta” and that there is no stable release channel yet.
Monthly compute cost of a hosted Buzz remote agent at the $0.65 per vCPU-hour rate on the buzz.xyz pricing page, for three usage patterns we assumed. Our arithmetic; data and inference charges are excluded.
What Was Posted on X
buzz 1.0 is finally here! the most flexible team workspace out there. for humans and agents, model agnostic, local or cloud, and infinitely extendable.
Jack Dorsey announced Buzz 1.0 on X at 02:38 UTC on 11 November, calling it the most flexible team workspace out there, for humans and agents, model agnostic, local or cloud, and infinitely extendable, with a link to buzz.xyz. The post does not mention Block or prices; this piece takes those from Buzz’s pricing page, its GitHub repositories and Block’s engineering blog.
Why Block Built It
Block introduced Buzz on its engineering blog on 21 July. The post says “The bottleneck moved from intelligence to coordination” and that “Agents made individual work faster and teamwork slower”.
Buzz is built on Nostr, an open protocol for signed messages, the post says, and “Buzz gives each agent its own key”, so an agent signs its own work instead of borrowing a person’s credentials. On open source, it says: “software got cheap. Taste didnโt”.
The Pricing Model
The Buzz homepage opens with “Goodbye Slack. Goodbye GitHub.” The pricing page has three tiers: a free self-hosted option using the open-source software, a pay-as-you-go hosted plan with no payment method needed to start, and an Enterprise plan with a custom agreement, SSO and dedicated tenancy.
The pay-as-you-go plan includes 100 GB of shared data per month and up to 10 static and 10 dynamic apps. Managed inference is billed at model-specific rates, and hosted apps are not billed for compute at launch, the page says. The rates it lists are for data, compute and inference, not per user. For Block’s own spending on software, see our piece on its latest filing.
Our Numbers
At $0.65 per vCPU-hour, an agent on one vCPU for a 176-hour working month of 8 hours a day over 22 days costs $114.40, and one on four vCPUs running all 730 hours costs $1,898, by our arithmetic. Both exclude data and inference charges.
The original block/buzz repository, created on 6 March, had 36,917 GitHub stars and 4,907 forks when we checked at 09:44 UTC on 11 November. That is about 148 stars a day over 250 days, by our arithmetic. The newer block/buzz-app repository had 153 stars.
The Structural Read
As we read the pricing page, the business lever is usage rather than headcount: it lists rates for data, compute and inference and no per-user fee, so a team’s bill would follow how long its agents run, not how many people join.
As we read Block’s blog post, identity is the design choice that matters most: each agent gets its own key and signs its own work, which the post presents as an alternative to giving agents a person’s credentials.
Block Engineering, blog post introducing Buzz, 21 July 2026
“The bottleneck moved from intelligence to coordination”
Three Implications
FOR TEAMS USING AGENTS The README says agents run on your computer while Buzz is open, with Claude, Codex, Goose and Pi as harnesses.
FOR SLACK AND GITHUB The Buzz homepage opens with “Goodbye Slack. Goodbye GitHub.” and pitches chat, planning, code and PRs in one place.
FOR SELF-HOSTERS Block’s engineering blog says the relay is a single Rust binary that needs Postgres, Redis and an S3-compatible object store.
The framework argues: “Close the SCARCE layer (keep proprietary) + Open the ABUNDANT layer (commoditize)”.
As we read it, Buzz applies that split: the workspace software and protocol are open, while the hosted plan charges for compute, data and inference. The framework predates Buzz and does not discuss it.
GitHub stars and forks for Block’s two Buzz repositories at 09:44 UTC on 11 November 2026, from the GitHub API. block/buzz was created on 6 March 2026; the 250-day span is our arithmetic.
What Is Not Established
The pages and posts we read give no count of Buzz users or paying workspaces, and no per-model inference rates. Dorsey’s post does not mention Block; the link between Buzz and Block comes from the block GitHub organisation and Block’s engineering blog.
The 730-hour and 176-hour months are our assumptions for illustration, not Buzz’s. For an earlier Block AI project, see our piece on Block’s voice AI.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Block’s Buzz 1.0 puts people and agents from Claude, Codex, Goose and Pi in one workspace, with the software free to self-host under Apache 2.0. The hosted service charges for usage, at $0.65 per vCPU-hour for remote agents, rather than a listed per-user fee.
A note on sourcing. We read buzz.xyz and its pricing page, the Buzz 1.0.0 release and the READMEs of Block’s block/buzz-app and block/buzz repositories on GitHub, GitHub’s repository data at 09:44 UTC on 11 November, and Block’s engineering blog post introducing Buzz. Monthly costs and rates per day are our arithmetic on usage patterns we assumed. Block was not contacted for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
